Purpose <p>The purpose of this research is to address the formulation and manufacturing challenges in pharmaceutical pellet production by integrating artificial intelligence (AI) techniques. The study aims to enhance the efficiency and quality of the extrusion-spheronization process, ultimately improving the overall production of pharmaceutical pellets.</p> Methods <p>This research comprised three phases. First, a comprehensive review of literature, industry handbooks, and prior studies identified key formulation and process variables impacting pellet quality in extrusion-spheronization, and gathered data from 158 pharmaceutical pellets. Second, diverse machine learning models (K-Nearest Neighbor, logistic regression, decision trees, random forest, XGBoost, deep classifier, and ensemble voting) classified pellet quality (high/low) based on a novel composite metric. High accuracy (&gt; 95%) achieved by deep learning and XGBoost informed feature importance analysis. Explainability techniques (Forward Selection, SHAP, and LIME) elucidated the influence of process variables and excipients. Frequent pattern mining (FP-Growth) and rule induction revealed significant association rules, offering insights into formulation-process interactions. Finally, a decision tree was constructed to generate actionable process guidelines for real-time decision support.</p> Results <p>The models like random forest, XGBoost and specialy deep learning demonstrated higher performance with accuracy exceeding 95% and F-measures above 97%. SHAP and LIME explanations identified critical process parameters—including extrusion speed, spheronization speed, and drying temperature—that significantly influence quality outcomes. Additionally, frequent pattern mining uncovered key association rules, which, along with decision tree models achieving approximately 88.47% accuracy, formed the basis for a practical production guideline integrated into an operational dashboard. The framework effectively supports real-time decision-making, quality control, and process optimization, with promising potential for further enhancement through larger, real-time datasets.</p> Conclusion <p>This study underscores the effectiveness of integrating machine learning, explainability techniques, and data mining to optimize pharmaceutical pellet manufacturing. The high-performing models and derived process rules offer a robust foundation for improving quality consistency and operational efficiency. The approach facilitates accessible, data-driven decision support, fostering continuous process improvement, with future prospects involving real-time data integration to further elevate predictive accuracy and manufacturing resilience.</p>